Genetic Algorithm Based Parameter Tuning for Robust Control of Launch Vehicle in Atmospheric Flight

Abstract

A hybrid approach to the design of the attitude control system for a launch vehicle (LV) in the atmospheric flight phase is proposed in this paper, where a structured $\mathcal {H}_\infty $ controller is tuned using a genetic algorithm (GA). The $\mathcal {H}_\infty $ synthesis relies on a classical architecture for the thrust vector control (TVC) system that features proportional-derivative loops and bending filters. Once a set of requirements on stability and robustness typical of industrial practice is specified, control design is carried out by parameterizing the $\mathcal {H}_\infty $ weighting functions, and solving a two-layer max-min global optimization problem for the tuning parameters. The design methodology is applied to the model of a medium-size LV. The novel design is analyzed in off-nominal conditions taking into consideration model parameter scattering and wind disturbances. The results show that the automated design procedure allows to devise time-scheduled controllers providing adequate stability and performance, and appears as a viable and effective solution in order to reduce the burden of recurrent activities for controller tuning and validation conducted prior to each launch.